Memory problems with script runner

[Hugo Mora]

Hi!

We are getting a strange error on our JIRA instance. JIRA is creating a very large object that crashes JIRA instance after GC timeout... 

We have detect a lot of threads with this stack trace.

Could you please review this error?

Thanks a lot,

--Hugo

http-bio-8080-exec-85
  at java.lang.StringCoding$StringDecoder.decode([BII)[C (Unknown Source)
  (...)
  at com.onresolve.jira.groovy.jql.AbstractScriptedJqlFunction.getIssues(Ljava/lang/String;Lorg/apache/lucene/search/Query;)Ljava/util/List; (AbstractScriptedJqlFunction.groovy:74)
  at sun.reflect.GeneratedMethodAccessor2154.invoke(Ljava/lang/Object;[Ljava/lang/Object;)Ljava/lang/Object; (Unknown Source)
  at sun.reflect.DelegatingMethodAccessorImpl.invoke(Ljava/lang/Object;[Ljava/lang/Object;)Ljava/lang/Object; (Unknown Source)
  at java.lang.reflect.Method.invoke(Ljava/lang/Object;[Ljava/lang/Object;)Ljava/lang/Object; (Unknown Source)
  at org.codehaus.groovy.runtime.callsite.PogoMetaMethodSite$PogoCachedMethodSite.invoke(Ljava/lang/Object;[Ljava/lang/Object;)Ljava/lang/Object; (PogoMetaMethodSite.java:226)
  at org.codehaus.groovy.runtime.callsite.PogoMetaMethodSite.callCurrent(Lgroovy/lang/GroovyObject;[Ljava/lang/Object;)Ljava/lang/Object; (PogoMetaMethodSite.java:52)
  at org.codehaus.groovy.runtime.callsite.AbstractCallSite.callCurrent(Lgroovy/lang/GroovyObject;Ljava/lang/Object;Ljava/lang/Object;)Ljava/lang/Object; (AbstractCallSite.java:145)
  at com.onresolve.jira.groovy.jql.AbstractScriptedJqlFunction.getIssues(Ljava/lang/String;)Ljava/util/List; (Unknown Source)
  at com.onresolve.jira.groovy.jql.AbstractScriptedJqlFunction$getIssues.callCurrent(Lgroovy/lang/GroovyObject;Ljava/lang/Object;)Ljava/lang/Object; (Unknown Source)
  at com.onresolve.jira.groovy.jql.EntityMatch.getQuery(Lcom/atlassian/jira/jql/query/QueryCreationContext;Lcom/atlassian/query/operand/FunctionOperand;Lcom/atlassian/query/clause/TerminalClause;)Lorg/apache/lucene/search/Query; (script14333834963861257966674.groovy:88)
  at com.onresolve.jira.groovy.jql.EntityMatch$getQuery.call(Ljava/lang/Object;Ljava/lang/Object;Ljava/lang/Object;Ljava/lang/Object;)Ljava/lang/Object; (Unknown Source)
  at com.onresolve.jira.groovy.jql.ScriptedJqlFunction.getQuery(Lcom/atlassian/jira/jql/query/QueryCreationContext;Lcom/atlassian/query/operand/FunctionOperand;Lcom/atlassian/query/clause/TerminalClause;)Lorg/apache/lucene/search/Query; (ScriptedJqlFunction.groovy:131)
  at com.onresolve.jira.groovy.jql.ScriptedJqlFunction$getQuery$0.call(Ljava/lang/Object;Ljava/lang/Object;Ljava/lang/Object;Ljava/lang/Object;)Ljava/lang/Object; (Unknown Source)
  at com.onresolve.jira.groovy.jql.ScriptedFunctionClauseFactory.getQuery(Lcom/atlassian/jira/jql/query/QueryCreationContext;Lcom/atlassian/query/clause/TerminalClause;)Lcom/atlassian/jira/jql/query/QueryFactoryResult; (ScriptedFunctionClauseFactory.groovy:52)

[Hugo Mora]

Hi!

More info about this issue

We have reproduce the error on a side environment, using a JQL like this:

{code}

project = PROYECT1 AND "CustomFieldA" in (VALUE1,VALUE2, VALUE3) AND issueFunction in issueFieldMatch("", labels, "1.61.0") ORDER BY updated DESC

{code}

We have desactivated module "Scripted JQL Function" (scripted-jql-function) and there's no fail at this moment.

Cheers,

--Hugo

[Nic Brough -Adaptavist-]

That would make sense - the "" in the issueFieldMatch means "read every issue in JIRA" - that's going to be a disaster in terms of performance in any medium to large JIRA installation.

[MattS]

You should be able to get this to work by replacing "" with a better restricting JQL query

[Nic Brough -Adaptavist-]

Script-runner itself will not cause memory errors (unless you've got too little memory to load it in the first place, but that would apply with any addon).  It's what you do with it that matters.

You'll need to take a close look at the query that your users are triggering to work out what it is doing here.

[Hugo Mora]

Hi Nic, Heap is setup with 4 GB. I think that we have located the function on the JQL. I will update this issue. Thanks, --Hugo